Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
INFORMATION PROCESSING DEVICE
Document Type and Number:
WIPO Patent Application WO/2019/065703
Kind Code:
A1
Abstract:
The purpose of the present invention is to improve identification accuracy. In the present invention, an image recognition device 200 comprises: an image processing device 21 that acquires a feature amount from an image; and an identification device 201 that determines, using the acquired feature amount, whether a prescribed identification subject is present in the image, and identifies the identification subject. The identification device 201 comprises a BNN that has learned the identification subject in advance, and performs identification processing by performing a binary calculation with the BNN on the feature amount acquired by the image processing device 21. At that time, the identification device 201 selects a portion effective for identification from among high-dimensional feature amounts output by the image processing device 21 to reduce the dimensions used in identification processing, and copies low-dimensional feature amounts output by the image processing device 21 to increase the dimensions. By selecting and copying feature amount dimensions, the feature amount dimensions used in identification can be appropriately adjusted while a demanded identification accuracy is ensured, so that the identification device 201 can be mounted on a small, low power-consumption hardware circuit.

Inventors:
YAMADA HIDEO (JP)
MURAMATSU RYUYA (JP)
SHIBATA MASATOSHI (JP)
TAMUKOH HAKARU (JP)
ENOKIDA SHUICHI (JP)
YAMASAKI YUTA (JP)
Application Number:
PCT/JP2018/035608
Publication Date:
April 04, 2019
Filing Date:
September 26, 2018
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
EQUOS RES CO LTD (JP)
KYUSHU INST TECH (JP)
International Classes:
G06T7/00; G06N3/04; G06N3/08; G06V10/50; G06V10/764
Foreign References:
JP2015014819A2015-01-22
Other References:
SHIGERU YAMAGUCHI ET AL.: "Fundamental description of a hybrid pattern recognition system using microlens array", December 1991 (1991-12-01), pages 1 - 7, XP055586624, Retrieved from the Internet
AKEDO YU ET AL: "A learning algorithm of Binary neural networks based on real-coded GA ", IEICE TECHNICAL REPORT, vol. 108, no. 101, 19 June 2008 (2008-06-19), pages 1 - 2, XP055681141, ISSN: 0913-5685
M. COURBARIAUX ET AL., BINARY CONNECT: TRAINING DEEP NEURAL NETWORKS WITH BINARY WEIGHTS DURING PROPAGATIONS, 18 April 2016 (2016-04-18), XP055447419, Retrieved from the Internet
MATTHIEU COURBARIAUX ET AL: "Binarized Neural Networks: Training Neural Networks with Weights and Activations Constrained to +1 or -1", 17 March 2016 (2016-03-17), pages 1 - 11, XP055405835, Retrieved from the Internet
TOMOKI WATANABESATOSHI ITO: "Co-occurrence Histograms of Oriented Gradients for Human Detection", IPSJ TRANSACTIONS ON COMPUTER VISION AND APPLICATIONS, vol. 2, 2010, pages 39 - 47
NAVNEET DALALBILL TRIGGS: "Histgrams of Oriented Gradients for Human Detection", IEEE COMPUTER SOCIETY CONFERENCE ON COMPUTER VISION PATTERN RECOGNITION, vol. 1, 2005, pages 886 - 893
See also references of EP 3690804A4
Attorney, Agent or Firm:
NAKANO Hitoshi et al. (JP)
Download PDF: